Data Classification via a New Data Mining Approach: Multiple Criteria Programming with Multiple Kernels

Nowadays the datasets are constituted by many high dimensional vectors in many applications. Due to this situation, it is necessary for these methods to reduce the computational burden and improve the generalization. In this paper, we propose a novel multiple criteria programming with multiple kernels (MK-MCP) to overcome these difficulties. This model adopts the equality constraints. Compared to the traditional mathematic programming based methods, this method doesnpsilat require solving a large scale formulation to find the optimal solution. So this formulation leads to an extremely fast and simple algorithm for generating a classifier that merely requires the solution of a set of linear equations. Inspired by the idea of SVM, we introduce the kernel function to solve the classification problem to improve the generalization ability. Some UCI datasets are used to demonstrate the efficiency of this model.